The polymorphic cytochrome P450 CYP2D6 catalyses the biotransformation of at least 40 drugs. The CYP2D6 genetic polymorphism is responsible for pronounced interindividual differences in plasma concentrations and, hence, in drug action and side-effects after administration of the same dose. Provided there is a close relationship between CYP2D6 genotypes and catalytic function, genotyping could be used in the clinical setting for individualization of drug dose. In the present study, we evaluated the relationship between the in-vivo enzyme activity and 35 different genotypes in order to determine whether genotyping can be used to predict a person's metabolic capacity for CYP2D6-catalysed drug oxidation using sparteine as a probe drug. One hundred and ninety-five Caucasian individuals were genotyped for seven nonfunctional (CYP2D6 x 3, x 4, x 5, x 6, x 7, x 8, x 16) and eight functional alleles (CYP2D6 x 1, x 2, x 2 x 2, x 2B, x 2B x 2, x 9, x 10, x 17). The metabolic ratio distribution for sparteine showed trimodality, with 15 poor metabolizers, 21 intermediate metabolizers, and 1.59 extensive and ultrarapid metabolizers. All poor metabolizers were unambiguously identified as carriers of two nonfunctional alleles. In contrast, the most frequent functional genotypes extensively overlapped and, with few exceptions, genotype was not a useful predictor of function. Gene dose effects among homozygotes and heterozygotes of the major functional alleles were not significant and could not explain the wide variations. Only a minor fraction of phenotypical ultrarapid metabolizers, arbitrarily defined as individuals with a metabolic ratio < 0.2, could be identified as carriers of three functional gene copies, including duplicated CYP2D6 x 2 x 2 alleles. Similarly, only a minor fraction of the intermediate metabolizers had predictive genotypes involving alleles coding for enzyme with impaired function. Thus, genotyping correctly identifies poor metabolizers, but quantitative prediction of drug metabolism capacity among extensive metabolizers is not possible.